The Opportunities for AI in education

Since its launch, ChatGPT has experienced widespread global uptake and sustained media attention. Dr Alfred Tiley of The Data Lab Innovation Centre explores its potential in education.

OpenAI’s ChatGPT is a powerful artificial intelligence (AI)-based chatbot that engages in human-like conversation with users via written text. It was released to the public in November 2022. Since its launch it has experienced widespread global uptake and sustained media attention.

ChatGPT performs very well in many complex tasks including writing code, composing essays, and providing personal tutoring. It has also been shown to achieve excellent results over a wide range of academic exams. Given its capabilities, and with the recent launch of rival chatbots like Google’s Bard and the power of future models set to increase, there are reportedly growing concerns among educators about the potential impact of the technology on the education sector.

So, are AI tools like ChatGPT valuable time-saving aides for students and teachers, or rather new and hard-to-detect means of plagiarism that are set to degrade independent thought and facilitate an overdependence on technology?

In March, I chaired a The Data Lab (TDL) event that brought together educators for an online discussion on the impacts of generative AI tools like ChatGPT in their field. I was joined by teachers, lecturers, and education experts, as well as data scientists and data engineers. Contrary to the alarm raised in some media reports, what emerged from the educators was a calm and lucid view of the risks posed by the new technology. Far from panicking, those present clearly outlined several strategies for how best to respond to – and embrace – the increased adoption of ChatGPT and AI across their sector.

Adapting to new AI

Apart from calls for a coherent set of institutional policies and education-specific government regulations on AI use, one of the most pressing concerns highlighted by attendees was the need to modernise an outdated assessment system; new AI tools like ChatGPT could be used by students to circumvent assessment procedures, particularly coursework or written exams. Several solutions to this issue were put forward including increased use of oral exams to tests students’ knowledge and critical thinking and allowing controlled use of ChatGPT and other AI tools – for example in some invigilated exams or in coursework if referenced.

Increased continual assessment was also proposed, where regular contact and greater involvement in students’ working processes allows teachers to detect sudden changes in the quality or style of their outputs – an indicator of potential AI use.

Despite a recent focus on technological solutions like AI plagiarism detectors, there was general agreement that they couldn’t be relied on to solve the problem alone. Fittingly, given that plagiarism and cheating are fundamentally human problems, the most potent solutions suggested were firmly human-based.

Embracing new AI

Attendees agreed that banning AI tools like ChatGPT could represent a failure to prepare students for a future workforce in which their use is widespread; students should be AI literate to make best use of the technology and to understand its limitations.

Despite reported concerns that ChatGPT could degrade students’ capacity for independent thought, there was a positive discussion around its use as an aide to foster critical thinking. For example, students could be asked to critique ChatGPT-generated writing. It was emphasised that embracing AI should not mean that fundamentals like reading, writing, arithmetic, and independent thought are neglected – these are enduring skills that will always be required alongside any technological advancements.

On a final optimistic note, attendees highlighted the potential for AI tools like ChatGPT to act as effective levellers in education, providing tailored assistance to students with additional educational needs and assisting busy teachers with heavy workloads – all for free (although this may not remain the case, as pointed out by Dr Gina Helfrich in The Data Lab’s ChatGPT Q&A blog). Tellingly, these points were also highlighted by the UK prime minister speaking at the London Tech Week in June.

As the event chair I was impressed by the roundly measured and pragmatic response from educators. The development of advanced AI chatbots and the technology that underpins them is unlikely to abate. Many other sectors are therefore likely to face similar AI-related challenges in parallel to those discussed here. Experts in those fields might take comfort from the robust and insightful discussion facilitated by this TDL event.